Climate action on the mind: a longitudinal study of Americans’ psychosocial drivers of high-impact climate behaviors

Abstract

Widespread adoption of high-impact climate-positive behaviors can significantly reduce greenhouse gas emissions. To motivate these behaviors, social scientists and policymakers need to understand people’s psychological and social (psychosocial) factors to create an environment that encourages widespread adoption. Past longitudinal survey research has focused on tracking changes in broad climate change beliefs and attitudes, risk perceptions, and climate policy support. While behavioral and psychological research has identified key beliefs and attitudes as enabling conditions, this work tends to study a single snapshot in time, often in a narrow population, not allowing for the identification of trends. In the present paper, we launch the annual Climate Action on the Mind (CAM) longitudinal survey, which aims to track key psychosocial factors shown to be important enablers of climate behaviors. Our work focuses on behaviors relevant to households in the United States, such as installing solar panels and driving an electric vehicle. The paper introduces the first two waves of the CAM survey from December 2021 (n = 2031) and June 2023 (n = 1528), quota-matched to represent the US adult population on key demographics. Our research offers novel insights into how the enabling psychological conditions for high-impact climate-positive behaviors are shifting in the United States, helping to inform the development of future communication strategies, interventions, and climate policy.

Introduction

At the current pace of carbon emissions, the world will burn through its target of limiting warming to 1.5°C by 2030. The significant adverse effects of climate change are becoming clear and dramatic every year [1].

Successfully addressing climate change requires comprehensive policy and regulatory initiatives at national and international levels. These policies are often not an alternative to behavior change, but rather mechanisms for driving shifts in behavior. For example, the Inflation Reduction Act—the most significant climate legislation in US history with the potential to cut the nation’s emissions by over 40% by 2035 [2]—aims to accelerate the adoption of electric vehicles, the installation of rooftop solar, adoption of efficient heating and cooling, among other shifts in household behavior. Given the sizeable carbon-mitigating potential of household behavior adoption at scale, future policies may further encourage behavior change [3]. The success of broad-spectrum climate policies hinges on not only targeting the right behaviors, in terms of carbon impact but also implementing policies and programs likely to have the greatest impact on their adoption. As such, social scientists, policymakers, and behavior change program designers must reliably identify beliefs and attitudes driving climate change through climate behaviors [4]. To inform the development of behavior-change campaigns and support tailored communication and intervention strategies to increase the adoption of climate behaviors it is critical to understand the prevalence of those beliefs and attitudes and how they are changing over time.

A review of key psychosocial factors motivating climate behaviors

Previous longitudinal polling in the USA examined trends in climate policy support [5, 6], risk perceptions [7, 8], and broad beliefs around climate change, such as the belief that it is happening [9]. While this research makes important contributions to helping understand how public opinion is evolving, it may not capture the specific beliefs and attitudes most connected with taking action.

Parallel to this, there has been substantial work in identifying key psychological and social (psychosocial) factors that motivate pro-environmental behaviors. Since psychological factors are more susceptible to targeted behavioral interventions than traditional demographic constructs or climate change beliefs [10], it is essential to integrate these psychosocial insights into climate policymaking and intervention design. By doing so, policymakers and behavior change program designers can develop more effective solutions that not only raise awareness but also actively encourage pro-environmental actions among the public. Our literature review identified the following set of factors, which, while not exhaustive, are found to be particularly relevant for driving the adoption of climate behaviors.

Social norms

Social norms refer to the informal rules and standards that guide behavior in groups and societies [11]. When a group widely accepts and endorses a behavior, its members are more likely to act in the same way [12]. Social norms are constituted by three distinct beliefs: empirical expectations (descriptive norms), personal normative beliefs, and normative expectations (injunctive norms) [11].

Empirical expectations are the beliefs about what others in an individual’s reference network are doing regarding climate action. When people believe that others in their social group are taking action, they are more likely to do so themselves [13]. Personal normative beliefs refer to an individual’s sense of moral obligation or duty [14, 15]. If individuals believe that it is their responsibility to act against climate change, they are more likely to do so. This sense of duty can stem from various sources, including religious, ethical, or moral convictions. Normative expectations are individuals’ beliefs about what others think they should do [11]. If people believe that their peers (e.g. friends, family members, or neighbors) expect them to adopt a climate behavior, they are more likely to do so themselves.

Social norms have been shown to influence people’s interest in various climate-positive behaviors, including installing solar panels [16], driving an electric vehicle [17], eating less meat [18], and reducing food waste [19].

Self-efficacy

Self-efficacy refers to an individual's belief in their capabilities to organize and execute the courses of action required to achieve certain goals [20, 21]. Confidence in one’s ability to change is a key determinant of climate-adaptive behaviors [22]. People with high self-efficacy—those who believe they have the knowledge, skills, and abilities to affect change—are more likely to adopt climate behaviors. Past research noted that increased self-efficacy via experience and knowledge holds the potential to increase the adoption of both electric vehicles [23] and solar panels [24].

Consideration

Consideration—thinking about a behavior, considering change, or evaluating consequences of change—is another, and often the first, important step of behavior change. In this early behavior change stage, people become aware of a problem (e.g. climate change) and start thinking about ways to address it, often weighing the positives and negatives of changing their behavior [25]. This stage is critical as it sets the foundation for subsequent stages of change. For example, an individual may become aware of the negative climate impacts of excessive driving and begin contemplating changes to their daily commute.

Perceived individual benefits

Perceived benefits—If individuals believe that acting against climate change will have personal benefits, they are more likely to take that action [26]. These benefits can be direct (e.g. saving money from added energy efficiency) or indirect (e.g. the perceived health benefits of cleaner air). For example, Solarize campaigns in Connecticut, which aim to accelerate the adoption of rooftop solar, revealed that households are more responsive to messages that focus on individual benefits than those that focus on community benefits [27].

Additional psychological influences

Social science literature also surfaces psychological properties of climate change itself that can influence pro-environmental intentions and behaviors. For instance, risk perceptions—how individuals perceive the risks associated with climate change—can drive climate action and were found to be an important predictor of public willingness to engage in mitigation behaviors [28, 29]. Similarly, climate change beliefs—including the belief in the existence and human causation of climate change—were found to be a predictor (albeit weak) of pro-environmental intentions and policy support [10]. Given our research focus on documenting and tracking behavior-specific beliefs, we did not assess general climate attitudes and beliefs or climate risk perceptions, which are regularly tracked in more general climate polling [30].

To increase behavioral and construct coverage, each construct was measured with a single item per behavior. Due to survey length limitations, we were significantly limited in the number of items we could include in our survey. We were therefore unable to track several other well-documented psychological influences, including perceived responsibility [22, 31, 32], perceived economic considerations beyond perceived benefits [33], and collective efficacy [34, 35], to name a few. Our selection of measures should therefore not be taken as an exhaustive list of the psychological states influencing climate-positive behaviors.

Upon review of the existing literature, a significant research gap becomes apparent. Despite the scientific understanding of general climate opinions and recognition of the key drivers of pro-environmental behaviors, there is limited research into how these determinants evolve on a large scale for behaviors that have the highest carbon mitigation potential. For instance, while we know that over 5.4 million electric cars have been sold in the United States to date [36] and that 36 GW of residential solar has been installed up to 2023 [37], we know very little about the prevalence of beliefs surrounding these behaviors, such as social norms, and even less about how these beliefs are changing over time. To accelerate the adoption of climate behaviors, it is important to collect large-scale data that begin to assess the acceptance of new behaviors. This behavioral data should evaluate different enablers or predictors of the public’s intention to act and give a glimpse of expected belief shifts in the population frequently.

Our research aims to bridge this gap. We chose to measure psychological properties of behaviors, such as empirical expectations (descriptive norms), normative expectations (injunctive norms), and self-efficacy as these measures correspond to three of the best-documented large drivers of climate-related behaviors (for meta-analysis, see [22]). We included a measure of perceived individual benefit of the behaviors, recognizing that this measure is a commonly supported driver for consumer behaviors [38]. Our 2023 wave and later waves additionally include measures of outcome efficacy, which was identified as a significant driver of climate-related behaviors [22]. Outcome efficacy estimates will be included in the future analyses of study waves. By focusing on these psychological factors and observing shifts over time, CAM allows for a more nuanced understanding of both barriers and facilitators of important climate behaviors.

We measured the prevalence of the key psychosocial factors identified above over two timeframes on a national level—2021 and 2023—for a set of high-impact climate behaviors. In addition, we measured reported adoption of the behaviors and intention to engage in the behaviors. We intend to collect these data on an annual basis to monitor and report potential changes in beliefs, intentions, and behaviors over time. This research is key to the home institution’s data for decision-making strategy, resulting in a high degree of institutional support. To financially support this effort, we have secured funding for the third annual round of data collection, analysis, and dissemination and have incorporated future rounds into the home institution’s annual fundraising pipeline going forward.

Materials and methods

Participants

We fielded the first wave in December 2021 (n = 2031) and the second in June 2023 (n = 1528) on the Luc.id Marketplace platform. Both waves were designed and analyzed to be representative of the US adult population on demographics, including age group, sex, ethnicity, and Hispanic/non-Hispanic origin. When reporting population estimates, we employed survey weights to ensure that the responses were representative of the actual population proportions. The demographic composition of respondents and US population estimates by demographics are available in Table 1.

We included multiple attention checks throughout (e.g. “This question makes sure you are paying attention. Please select seven on this question.”) to ensure data quality [39]. Participants that failed an attention check were disqualified from continuing the study and their partial response was not included in our analyses. Past research with respondents from similar online platforms has found that resulting demographic and empirical findings track well with US national probability sample benchmarks [40]. Before participating, all respondents provided their consent.

Procedure and measures

Across the two waves, we studied three high-impact climate-mitigating behaviors, allowing us to estimate potential shifts in psychosocial enablers for each: installation of solar panels, purchase of an Electric Vehicle (EV), and purchase of carbon offsets (we note that carbon credits have been criticized for their mitigation potential, particularly in the context of forest conservation [41]. However, others have argued that well-designed carbon offset projects, for example sequestration in aquatic environments, can still serve an important role in climate change mitigation and adaptation [42, 43]. It is critical that any carbon offset project promoted to the public be vetted in terms of permanence, leakage, and additionality [44]).

These behaviors were selected due to their relevance in the US context, cultural feasibility, regulatory context, and moderate to high carbon-mitigating potential, as identified by prior work [45]. This work identified six behaviors from a list of fifty-five potential actions and modeled potential emissions reductions using historical data and assumptions for adoption rates (for more details about the process of selecting climate behaviors and impact modeling, see [45]). Between the two waves of the CAM survey, we examined psychological mechanisms related to all six behaviors identified as important for policymakers to promote for climate mitigation. Additionally, our choice of behaviors closely aligns with the set of most impactful household climate actions identified by prior work [3, 46]. In addition to the three behaviors present in both waves, we surveyed an additional eight climate behaviors (ranging from eating less meat to installing a heat pump for heating and cooling), for which we have single-point-in-time estimates (see Supplementary material). While our choice of behaviors focuses on high-impact climate actions that people in the United States could feasibly adopt, we recognize that there are other solutions that have smaller, but still important, impact at different scales or locations.

For each climate behavior, study participants responded to measures identified in the literature as key motivators of behavior change. For each enabler, Table 2 presents example survey questions with response ranges.

In addition to the predictors of climate action outlined above, we measured the reported adoption of the behaviors (e.g. “Does your household have solar panels installed to get electricity?”—Yes or No).

To construct survey weights and conduct exploratory analyses, we collected several demographic measures, including gender, age, race, level of education completed, annual household income, geography of residence, and political viewpoint. Full survey instruments, including survey questions about other behaviors, and the data underlying this research are available in the public OSF repository (https://osf.io/tk9j4/).

Results

Figure 1 presents the changes in the psychosocial enablers over time. For additional results, see the Supplementary material.

Installation of solar panels

Our results indicate a statistically significant increase in the average intention to install solar panels (M2021 = 0.16, M2023 = 0.21; t(3557) = −3.90; Cohen’s d = −0.57; P < .05). In 2021, the federal tax credit for solar panel installation was 26%, and it was increased to 30% in 2022 under the Inflation Reduction Act [47]. A survey of US homeowners in 2022 revealed that 64% of people cited solar investment tax credits as a reason they have installed or have considered installing solar panels [48]. While the present study is not designed to explain why reported intention increased, we hypothesize that renewal and extension of the federal tax credit played a role in increasing intention to adopt solar energy.

We saw a statistically significant increase in two social norms: personal normative beliefs and normative expectations. In 2023, more people believed that installing solar is the right thing to do (M2021 = 0.61, M2023 = 0.63; t(3557) = −2.44; d = −0.97; P < .05) and people thought a greater number of their peers believe that people should install solar (M2021 = 0.32; M2023 = 0.34; t(3557) = −2.78; d = −0.62; P < .05). These beliefs can serve as early signals of how well the behavior aligns with the accepted standards and norms in society. An increase in these norms, while relatively small in terms of practical significance, indicates a growing normative acceptability of technology. Our findings are broadly aligned with polling data showing that most Americans—82% – support the expansion of solar power [49].

Results indicate no statistically significant change in empirical expectations: the perceived number of people who have installed solar panels did not change. However, recent homeowner polling [49] and solar industry data [50] indicate a growth in residential solar installations, suggesting a lag in people’s perception of the prevalence of solar in their community.

A decrease in self-efficacy (M2021 = 0.44, M2023 = 0.41; t(3557) = 3.61; d = 0.76; P < .05), personal benefit of installing solar (M2021 = 0.61, M2023 = 0.55; t(3557) = 4.33; d = 0.71; P < .05), and consideration (M2021 = 0.52, M2023 = 0.46; t(3557) = 3.59; d = 1.01; P < .05) could indicate a lack of sufficient effort in building confidence in the technology. For example, a dip in self-efficacy could indicate that educational and outreach campaigns are not effectively communicating how homeowners can transition to solar energy. If solar energy companies are not effectively marketing or if other energy sources dominate the conversation, solar might get sidelined. Since seeing neighbors or friends benefit from solar panels can motivate others to consider them [16], initiatives that promote group-buying discounts and visible success stories can bridge the confidence and consideration gaps. A second more troubling explanation could be that as solar becomes more popular, people are observing others having difficulty and not seeing the benefits promised by installers. This might suggest the spread of solar could also be calcifying the views of non-adopters that it is not appropriate for them. Future research is necessary to disentangle possible explanations to plot an effective path forward.

Driving an electric vehicle

Compared to 2021, in 2023 fewer people considered driving an electric car (M2021 = 0.51, M2023 = 0.41; t(2938) = 7.10; d = 1.00; P < .05); average intention to make the next car electric also fell (M2021 = 0.38, M2023 = 0.32; t(3086) = 5.37; d = 0.64; P < .05). Similarly, our results indicate a reduction in the perceived personal benefit of electric vehicles (M2021 = 0.50, M2023 = 0.41; t(3557) = 8.60; d = 0.74; P < .05) and a decline in the public’s confidence to own and drive one (M2021 = 0.52, M2023 = 0.46; t(3557) = 8.29; d = 0.72; P < .05). High inflation rates, fear of an economic downturn, and the potential of a recession may have shifted consumer priorities away from what is commonly perceived as a luxury purchase [51]. Additionally, supply chain issues may have affected the availability of electric vehicle models on the market, further impacting affordability.

We also observe a decrease in personal normative beliefs (M2021 = 0.51, M2023 = 0.40; t(3557) = 6.68; d = 1.00; P < .05) and normative expectations (M2021 = 0.30, M2023 = 0.27; t(3557) = 3.58; d = 0.59; P < .05). This suggests that fewer people believe purchasing an electric car is the right thing to do, and that they feel less societal pressure to drive one.

Conversely, there is a growing perception of the prevalence of electric cars on the road, as demonstrated by a statistically significant increase in empirical expectations (M2021 = 0.13, M2023 = 0.17; t(3557) = −4.27; d = −0.39; P < .05). This public perception aligns with industry data, which shows that the proportion of EVs on the road reached its highest level in 2022 [52]. An alternative explanation for these findings could be that exposure to EVs may counterintuitively be reducing interest. Other research has found that those considering EVs list a variety of structural barriers such as charging logistics, range on a single charge, and lack of desired models [53]. It is possible that, after seeing more EVs on the road people have learned about these difficulties, resolving what was previously ambiguous into resistance. If this were the case, it would bolster calls for greater investment in structural support for EVs to avoid non-actors cementing their views on EVs’ lack of suitability.

Purchase of carbon offsets

Our results indicate no statistically significant change in the share of the population that considered buying offsets, despite the growth of voluntary offset markets over time [54]. We saw no statistically significant differences in people’s intention to buy carbon offsets either, with the median response for buying carbon offsets at zero percent chance. It is likely that low consumer awareness of carbon credits and the perceived complexity of climate solutions are slowing down the intention [55]. Perhaps for the same reason, we observed a drop in the perceived personal benefit of buying carbon offsets (M2021 = 0.45, M2023 = 0.36; t(3557) = 6.31; d = 0.75; P < .05). There were no statistically significant movements in self-efficacy and the two social norms surrounding offset purchasing: personal normative beliefs and normative expectations.

Despite slow progress on key psychosocial enablers, we observed a notable rise in the number of people reporting their purchase of carbon offsets (M2021 = 0.03, M2023 = 0.06; t(3557) = −4.96; d = −0.42; P < .05) and an increase in the belief that others are purchasing carbon offsets (M2021 = 0.10, M2023 = 0.15; t(3557) = −4.67; d = −0.43; P < .05). It is possible that the different descriptions of carbon offsets across two studies might have influenced these reported changes (see Supplementary material). As the voluntary carbon offset market becomes more familiar to the American public, skepticism might arise regarding the actual impact and additionality of carbon offset projects. While solving for the lack of knowledge will remove some barriers to purchasing for consumers, increasing transparency is likely the key to increasing consideration and purchasing of offsets. In addition, solutions that remove friction in the purchasing process can bolster efficacy, in turn increasing the likelihood of adoption.

Discussion and implications

By focusing on key psychological enablers of climate action, the CAM survey helps better understand the present state of factors driving climate behaviors. Given the relevance of these factors, behavior change campaigns could benefit not only from communicating available incentives and the growing availability of solutions but also from improving perceived self-efficacy, highlighting personal benefits, and fostering normative responsibility and societal acceptance. CAM data complements existing tracking of population-level climate change beliefs by providing deeper insights into the current levels of the psychological states critical for the adoption of meaningful climate behaviors. By providing these data on a continued basis, we aim to encourage researchers and practitioners to collaborate and leverage these insights to create an enabling environment for climate behaviors, thus improving potential intervention strategies for addressing climate change.

Many of our findings stand in contrast to what one might assume if simply looking at the adoption trends of these behaviors over time. Adoption being on the rise across these behaviors might lead one to believe that we are in the hockey-stick phase, self-catalyzing adoption without any additional support. However, our tracking of the mental states enabling this adoption suggests that we are far from mass adoption. While adoption is rising, we are observing a reduction in many key enabling states. Each of our behaviors shows a reduction in perceived personal benefit, solar and EVs show a reduction in self-efficacy, and perhaps most concerningly, fewer people are considering EVs and solar in 2023 than were in 2021. The differences between consideration, reported adoption, and reported intention for the behaviors surveyed may be attributable to the intention-behavior gap. Past research has documented factors contributing to this gap, including perceived economic costs, lack of social or cultural norms, and insufficient infrastructural support [56]. We explored changes in psychological states across US regions (Supplementary Fig. S3). We saw a decline in psychosocial states for EVs across all regions, with some of the biggest declines in the South. Changes in psychosocial states for solar and carbon offsets show more variation, with increases in the Pacific and Mountain regions. The South and Southeast generally show decreases or minimal changes across all three behaviors. We also evaluated whether these enablers have become polarized over the years and found no consistent evidence for political polarization over time (Supplementary Fig. S2). See the Supplementary material for more details on these exploratory analyses.

A particularly concerning possible explanation for this pattern of results is that as individuals have greater social experience with these behaviors, they experience difficulties in their adoption of which they were previously unaware. This shift from unawareness to resistance could hinder the goal of widespread adoption if not addressed early in the socialization of these practices. These factors paint a possibly grimmer picture, where without policy and programming focusing on creating and enabling a psychosocial environment, we may observe a cap on the addressable market for these behaviors. Future work is needed to pin down the forces driving these changes in key psychosocial states which will support the precise targeting of future interventions to support the widespread adoption of these behaviors. Future waves of CAM will allow us to assess if these interventions as a whole are successful in creating the necessary psychosocial enabling environment to motivate climate behaviors.

The findings presented here should be interpreted in light of several limitations. Our research relied on self-reported data and while we aimed to reduce the impact of social-desirability bias by ensuring respondents knew their responses were anonymous, it may still well be present. Given survey length limitations, behaviors that we investigated represent only a subset of high-impact household actions to mitigate climate change. Future research could, and future waves of the CAM survey will, evaluate a wider array of behaviors. Similarly, while we aimed to examine key psychological constructs that existing literature has demonstrated as relevant for influencing climate action, there are several other psychological and social factors (for example, perceived responsibility to act or personal experience of climate change) that we did not capture in the present study. Finally, while the changes in the description of electric vehicles and carbon offsets were important for ensuring ongoing data quality, these changes may have influenced the responses. Future CAM survey iterations will continue to rely on the updated descriptions, allowing for consistent comparisons over time.

Acknowledgements

We would like to thank the reviewers of this manuscript, whose thoughtful feedback and recommendations helped us strengthen our work. We thank the study respondents who provided their time to support this research.

Author contributions

Rakhim Rakhimov (Data curation [lead], Formal analysis [equal], Project administration [lead], Visualization [lead], Writing—original draft [lead], Writing—review & editing [equal]), Scovia Aweko (Formal analysis [equal], Writing—original draft [supporting]), Erik Thulin (Conceptualization [lead], Funding acquisition [lead], Supervision [lead], Visualization [supporting], Writing—original draft [supporting], Writing—review & editing [supporting])

Supplementary data

Supplementary data is available at Oxford Open Climate Change online.

Conflict of interest: None declared.

Funding

This work was conducted with support from the Grantham Environmental Trust, the Gordon and Betty Moore Foundation, and the Arthur Vining Davis Foundations.

Data availability

Study data is publicly available: https://osf.io/tk9j4/.

References